Date Layout
Yanyu Liang
2017-03-28
Last updated: 2017-03-28
Code version: 096c6e3
Cell type and annotation type
Overview
As you can see, some cell types has multiple annotations (either different annotation or different experimental treatments). DNase and CTCF have data from multiple cell types. Also, some TFs, say Pol2, have data sets from the same cell type across multiple treatments.
library(ggplot2)
library(plyr)
data <- read.table('../data/aucs.txt', sep = '\t', header = T)
data$row_num <- seq(nrow(data))
ggplot(data) + geom_bin2d(aes(x = Cell.Type, y = TF.DNase.HistoneMark)) + theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title=element_text(size=22,face="bold"))
Cell type and annotation type pair that appears multiple times
This may due to multiple experimental treatments but some of them are duplicated experiments. The following lists all duplicated (cell type, annotation type) pair.
temp <- aggregate(data[, c('Treatment')], list(data$Cell.Type, data$TF.DNase.HistoneMark), function(x){
# print(x)
paste(x, collapse = ' <> ')
})
temp2 <- aggregate(data[, c('Treatment')], list(data$Cell.Type, data$TF.DNase.HistoneMark), function(x){
# print(x)
length(x)
})
temp3 <- aggregate(data[, c('row_num')], list(data$Cell.Type, data$TF.DNase.HistoneMark), function(x){
# print(x)
paste(x, collapse = ' <> ')
})
temp$num <- temp2$x
temp$row_num <- temp3$x
temp[temp$num > 1,] Group.1 Group.2 x num row_num
12 K562 ATF3 None <> None 2 334 <> 595
29 HepG2 BHLHE40 None <> None 2 297 <> 549
45 MCF10A-Er-Src c-Fos EtOH_0.01pct <> 4OHTAM_1uM_12hr <> 4OHTAM_1uM_4hr <> 4OHTAM_1uM_36hr 4 682 <> 683 <> 684 <> 685
50 K562 c-Jun IFNa30 <> IFNa6h <> IFNg30 <> IFNg6h <> None 5 606 <> 607 <> 608 <> 609 <> 610
52 H1-hESC c-Myc None <> None 2 473 <> 734
53 HeLa-S3 c-Myc None <> None 2 508 <> 737
56 K562 c-Myc IFNa30 <> IFNa6h <> IFNg30 <> IFNg6h <> None <> None <> None 7 611 <> 612 <> 613 <> 614 <> 615 <> 616 <> 746
57 MCF-7 c-Myc estrogen <> serum_stimulated_media <> serum_starved_media <> vehicle 4 749 <> 750 <> 751 <> 752
58 MCF10A-Er-Src c-Myc EtOH_0.01pct <> 4OHTAM_1uM_4hr 2 686 <> 687
66 HepG2 CEBPB None <> forskolin <> None 3 298 <> 551 <> 552
68 K562 CEBPB None <> None 2 338 <> 603
71 H1-hESC CHD1 None <> None 2 130 <> 470
81 K562 COREST None <> None 2 617 <> 618
84 A549 CTCF DEX_100nM <> EtOH_0.02pct <> None <> None 4 174 <> 175 <> 721 <> 764
106 GM12878 CTCF None <> None <> None <> None 4 128 <> 416 <> 727 <> 782
112 H1-hESC CTCF None <> None <> None 3 131 <> 263 <> 735
122 HeLa-S3 CTCF None <> None <> None 3 135 <> 738 <> 792
123 HepG2 CTCF None <> None <> None <> None 4 138 <> 300 <> 741 <> 793
127 HMEC CTCF None <> None 2 140 <> 797
135 HUVEC CTCF None <> None <> None 3 146 <> 744 <> 803
138 K562 CTCF None <> None <> None <> None <> None 5 150 <> 339 <> 619 <> 747 <> 805
139 MCF-7 CTCF estrogen <> serum_stimulated_media <> serum_starved_media <> None <> vehicle <> None 6 753 <> 754 <> 755 <> 756 <> 757 <> 806
144 NHEK CTCF None <> None <> None 3 165 <> 761 <> 809
145 NHLF CTCF None <> None 2 168 <> 810
150 SK-N-SH_RA CTCF None <> None 2 386 <> 813
201 HeLa-S3 DNase IFNa4h <> None 2 15 <> 56
238 Ishikawa DNase Estradiol_100nM_1hr <> 4OHTAM_20nM_72hr 2 23 <> 24
241 LNCaP DNase androgen <> None 2 25 <> 63
242 MCF-7 DNase Hypoxia_LacAcid <> None 2 26 <> 64
271 Urothelia DNase None <> UT189 2 39 <> 40
273 WI-38 DNase 4OHTAM_20nM_72hr <> None 2 124 <> 125
280 K562 E2F6 None <> None 2 341 <> 621
281 GM12878 EBF1 None <> None 2 212 <> 418
299 ECC-1 ERalpha BPA_100nM <> Estradiol_10nM <> Genistein_100nM 3 199 <> 200 <> 201
300 T-47D ERalpha BPA_100nM <> Genistein_100nM <> Estradiol_10nM 3 394 <> 395 <> 396
325 HepG2 FOXA1 None <> None 2 303 <> 304
341 K562 GATA2 None <> None 2 347 <> 624
343 MCF-7 GATA3 None <> None 2 696 <> 697
346 A549 GR DEX_500pM <> DEX_50nM <> DEX_5nM <> DEX_100nM 4 181 <> 182 <> 183 <> 184
462 K562 HDAC2 None <> None 2 153 <> 348
465 HepG2 HNF4A None <> forskolin 2 308 <> 558
471 K562 IRF1 IFNa30 <> IFNa6h <> IFNg30 <> IFNg6h 4 629 <> 630 <> 631 <> 632
477 H1-hESC JunD None <> None 2 268 <> 476
479 HepG2 JunD None <> None 2 310 <> 561
488 HepG2 MafK None <> None 2 563 <> 564
497 K562 Max None <> None 2 349 <> 637
547 HepG2 NRSF None <> None 2 314 <> 315
551 SK-N-SH NRSF None <> None 2 383 <> 384
554 GM12878 p300 None <> None <> None 3 225 <> 430 <> 431
557 HepG2 p300 None <> None 2 316 <> 569
558 K562 p300 None <> None 2 155 <> 645
571 A549 Pol2 DEX_100nM <> EtOH_0.02pct <> None 3 187 <> 188 <> 722
575 GM12878 Pol2 None <> None <> None <> None 4 231 <> 432 <> 434 <> 728
576 GM12891 Pol2 None <> None 2 250 <> 452
577 GM12892 Pol2 None <> None 2 257 <> 454
584 H1-hESC Pol2 None <> None 2 273 <> 736
587 HeLa-S3 Pol2 None <> None <> None 3 294 <> 529 <> 739
588 HepG2 Pol2 None <> forskolin <> None <> None 4 318 <> 571 <> 572 <> 742
589 HUVEC Pol2 None <> None <> None 3 333 <> 587 <> 745
591 K562 Pol2 None <> IFNa30 <> IFNa6h <> IFNg30 <> IFNg6h <> None <> None <> None 8 355 <> 646 <> 647 <> 648 <> 649 <> 650 <> 653 <> 748
592 MCF-7 Pol2 serum_stimulated_media <> serum_starved_media <> None 3 758 <> 759 <> 760
593 MCF10A-Er-Src Pol2 EtOH_0.01pct <> 4OHTAM_1uM_36hr 2 689 <> 690
618 K562 Pol2(phosphoS2) None <> None 2 651 <> 652
629 GM12878 Rad21 None <> None 2 234 <> 436
630 H1-hESC Rad21 None <> None 2 275 <> 481
632 HepG2 Rad21 None <> None 2 319 <> 574
634 K562 Rad21 None <> None 2 357 <> 655
650 K562 SETDB1 MNaseD <> None 2 658 <> 659
686 K562 STAT1 IFNa30 <> IFNa6h <> IFNg30 <> IFNg6h 4 662 <> 663 <> 664 <> 665
687 K562 STAT2 IFNa30 <> IFNa6h 2 666 <> 667
690 MCF10A-Er-Src STAT3 EtOH_0.01pct_4hr <> EtOH_0.01pct_12hr <> EtOH_0.01pct <> 4OHTAM_1uM_12hr <> 4OHTAM_1uM_36hr 5 691 <> 692 <> 693 <> 694 <> 695
709 K562 TBLR1 None <> None 2 669 <> 670
722 HeLa-S3 TCF7L2 None <> None 2 539 <> 540
739 A549 USF1 DEX_100nM <> EtOH_0.02pct <> EtOH_0.02pct 3 193 <> 194 <> 195
752 GM12878 YY1 None <> None 2 245 <> 447
758 K562 YY1 None <> None <> None 3 370 <> 371 <> 677
781 K562 ZNF274 None <> None 2 680 <> 681Same annotation type, different cell types
Remove all treatments other than “None” (namely ignore complexity from experimental treatment) and count the number of experiments we have per cell type and annotation type pair. Here, CTCF and DNase have many experiments across cell types.
data_none <- data[data$Treatment == 'None' & !is.na(data$Treatment),]
ggplot(data_none) + geom_bin2d(aes(x = Cell.Type, y = TF.DNase.HistoneMark)) + theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title=element_text(size=22,face="bold"))
Here we further explore the performance of DeepSEA on different cell types given a particular annotation type (only consider “None” treatment).
Performance distribution within a particular annotation type across cell types
Check out yanyu_lib.R here
source('../../cell_type/yanyu_lib.R')
data_none_ctcf <- data_none[data_none$TF.DNase.HistoneMark == 'CTCF',]
data_none_ctcf <- as_num(data_none_ctcf)
p1 <- ggplot(data_none_ctcf) + geom_histogram(aes(x = DeepSEA.ROC.AUC), bins = 20) + ggtitle('ROC AUC in CTCF data sets')
p2 <- ggplot(data_none_ctcf) + geom_histogram(aes(x = DeepSEA.PR.AUC), bins = 20) + ggtitle('PR AUC in CTCF data sets')
multiplot(p1, p2)
data_none_dnase <- data_none[data_none$TF.DNase.HistoneMark == 'DNase',]
data_none_dnase <- as_num(data_none_dnase)
p1 <- ggplot(data_none_dnase) + geom_histogram(aes(x = DeepSEA.ROC.AUC), bins = 20) + ggtitle('ROC AUC in DNase data sets')
p2 <- ggplot(data_none_dnase) + geom_histogram(aes(x = DeepSEA.PR.AUC), bins = 20) + ggtitle('PR AUC in DNase data sets')
multiplot(p1, p2)
As you can see, some cell types have low accuracy than others.
Distance between datasets
For each data set pair, compute how many instances have the same labels and how many have different labels. This setup is done by the following scripts at /project2/xinhe/yanyul/deep_variant/yanyu/DeepSEA/ and test_all.h5 is generated from test.mat by /project2/xinhe/yanyul/deep_variant/yanyu/DanQ/my_train/scripts/generate_test_data.py. The following shows how /data/test_all.h5.y_disance.hdf5 is generated. Check out compute_distance.py here
$ python my_scripts/compute_distance.py ../DanQ/my_train/data/test_all.h5 my_test/To compute the distance between two sets, Jaccard distance is used here (check definition here)
library(h5)
library(reshape2)
dnase_start <- 1
dnase_end <- 125
tf_start <- 126
tf_end <- 815
hist_start <- 816
hist_end <- 919 # these information is derived from /DeepVariantAnnotation/data/nmeth.3547-S3.xlsx
f <- h5file('../data/test_all.h5.y_disance.hdf5', 'r')
distance <- 1 - f['one_one'][] / (f['one_one'][] + f['zero_one'][] + f['one_zero'][])
melted_distance <- melt(distance)
ggplot(melted_distance) + geom_tile(aes(x = Var1, y = Var2, fill = value)) + scale_fill_gradient(low = "white", high = "black") + geom_vline(xintercept = (dnase_end + tf_start) / 2, color = 'red') + geom_hline(yintercept = (dnase_end + tf_start) / 2, color = 'red') + geom_vline(xintercept = (tf_end + hist_start) / 2, color = 'red') + geom_hline(yintercept = (tf_end + hist_start) / 2, color = 'red') + coord_fixed() + ggtitle('Distance between datasets \n (red line separate DNase/TF/Hist, southwest -> northeast)') + labs(x = 'dataset ID', y = 'dataset ID', fill = 'Jaccard \n distance')
The trend is within-group distance is smaller than between-group distance.
CTCF zoom in
Let’s only take a look at instances with CTCF binding (treatment = ‘None’). Most of them have some similarity to other members but there are two out liers (ID = 775, 815).
out <- zoom_in(melted_distance, data_none_ctcf)
sub.ctcf <- out$sub
data_none_ctcf <- out$ori
ggplot() + geom_tile(data = sub.ctcf,aes(x = as.factor(Var1), y = as.factor(Var2), fill = value)) + scale_fill_gradient(low = "white", high = "black") + coord_fixed() + theme(axis.text.x = element_text(angle = 90, hjust = 1), axis.text = element_text(size = 5)) + theme(legend.position = "bottom")
The following shows the relationship between mean distance within-group and ROC/PR AUC. As you can see, ID 755 and 815 are outliers.
p1 <- ggplot(data_none_ctcf, aes(x = dist.mean, y = DeepSEA.ROC.AUC, label = row_num)) + geom_text()
p2 <- ggplot(data_none_ctcf, aes(x = dist.mean, y = DeepSEA.PR.AUC, label = row_num)) + geom_text()
multiplot(p1, p2, cols=1)
For your interest, the following shows the cell types that either have worst ROC/PR AUC or are far from others.
Select the ones which are furthest to others.
data[data_none_ctcf[order(data_none_ctcf$dist.mean, decreasing = T)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
775 GM12801 CTCF None 0.996388547 0.9953458122 0.9970761167 0.262530792 0.3311145781 0.3156353128 775
815 WI-38 CTCF None 0.934407177 0.9415380569 0.9465819889 0.6887911683 0.7130838079 0.7255293238 815
796 HL-60 CTCF None 0.99534579 0.9964539538 0.9963882656 0.6719854829 0.7162271327 0.7213985799 796
735 H1-hESC CTCF None 0.986843822 0.9908473757 0.9916254659 0.5461145512 0.6243978119 0.6397933306 735
756 MCF-7 CTCF None 0.979445883 0.984418652 0.9851506937 0.5992568532 0.6440205591 0.652074639 756Select the ones which gets lowest ROC AUC scores.
data[data_none_ctcf[order(data_none_ctcf$DeepSEA.ROC.AUC)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
815 WI-38 CTCF None 0.934407177 0.9415380569 0.9465819889 0.6887911683 0.7130838079 0.7255293238 815
135 HeLa-S3 CTCF None 0.964978887 0.9710963274 0.9741564603 0.6521248673 0.6899951195 0.7035668817 135
170 Osteobl CTCF None 0.96530837 0.9720288862 0.974641306 0.6701567958 0.7027804547 0.7124233217 170
150 K562 CTCF None 0.970549293 0.9758263405 0.9778692369 0.6238341427 0.6608491882 0.6744054116 150
144 HSMMtube CTCF None 0.972644397 0.9779363134 0.9807862107 0.6377985839 0.6742370664 0.6878329335 144Select the ones which gets lowest PR AUC scores.
data[data_none_ctcf[order(data_none_ctcf$DeepSEA.PR.AUC)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
775 GM12801 CTCF None 0.996388547 0.9953458122 0.9970761167 0.262530792 0.3311145781 0.3156353128 775
732 GM19239 CTCF None 0.981206568 0.9854088462 0.9863621522 0.5228896013 0.5787069864 0.5869761929 732
735 H1-hESC CTCF None 0.986843822 0.9908473757 0.9916254659 0.5461145512 0.6243978119 0.6397933306 735
416 GM12878 CTCF None 0.979011087 0.9840308561 0.9850540848 0.5499689112 0.6069305147 0.6184375267 416
730 GM12892 CTCF None 0.976850556 0.9838173802 0.9840161027 0.5557373624 0.6069473908 0.6099822526 730DNase zoom in
The same analysis is done on DNase group (treatment = ‘None’).
out <- zoom_in(melted_distance, data_none_dnase)
sub.dnase <- out$sub
data_none_dnase <- out$ori
ggplot() + geom_tile(data = sub.dnase,aes(x = as.factor(Var1), y = as.factor(Var2), fill = value)) + scale_fill_gradient(low = "white", high = "black") + coord_fixed() + theme(axis.text.x = element_text(angle = 90, hjust = 1), axis.text = element_text(size = 5)) + theme(legend.position = "bottom")
p1 <- ggplot(data_none_dnase, aes(x = dist.mean, y = DeepSEA.ROC.AUC, label = row_num)) + geom_text()
p2 <- ggplot(data_none_dnase, aes(x = dist.mean, y = DeepSEA.PR.AUC, label = row_num)) + geom_text()
multiplot(p1, p2, cols=1)
Select the ones which are furthest to others.
data[data_none_dnase[order(data_none_dnase$dist.mean, decreasing = T)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
49 Caco-2 DNase None 0.955102255 0.9602247859 0.9614734632 0.4413681283 0.4660432973 0.4649057247 49
67 GM06990 DNase None 0.902955171 0.9063258167 0.9130242555 0.3256947324 0.3396407621 0.3519833383 67
122 Th2 DNase None 0.905005703 0.9103992578 0.9111198867 0.3293858306 0.3493359431 0.3472368336 122
121 SK-N-SH_RA DNase None 0.939063718 0.9438625773 0.9498986815 0.3975102061 0.4261801756 0.4398179731 121
16 Hepatocytes DNase None 0.881975064 0.8856014479 0.8869534689 0.285556575 0.3026511231 0.3029602655 16Select the ones which gets lowest ROC AUC scores.
data[data_none_dnase[order(data_none_dnase$DeepSEA.ROC.AUC)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
5 Fibrobl DNase None 0.837919632 0.8407263803 0.8431350521 0.3407087383 0.3526279102 0.3572207165 5
28 Melano DNase None 0.847363066 0.8506737609 0.8531173679 0.3536066271 0.3662819206 0.3742151035 28
30 Osteobl DNase None 0.849303246 0.8538104396 0.8558780011 0.3503940085 0.363640452 0.367974622 30
27 Medullo DNase None 0.864042242 0.8664238813 0.8692186275 0.3158031211 0.3271526972 0.3335229698 27
66 Th1 DNase None 0.865216236 0.8684588339 0.8701070274 0.3582683957 0.3696926502 0.3746784214 66Select the ones which gets lowest PR AUC scores.
data[data_none_dnase[order(data_none_dnase$DeepSEA.PR.AUC)[1:5],]$row_num,] Cell.Type TF.DNase.HistoneMark Treatment DeepSEA.ROC.AUC DanQ.ROC.AUC DanQ.JASPAR.ROC.AUC DeepSEA.PR.AUC DanQ.PR.AUC DanQ.JASPAR.PR.AUC row_num
16 Hepatocytes DNase None 0.881975064 0.8856014479 0.8869534689 0.285556575 0.3026511231 0.3029602655 16
74 HA-sp DNase None 0.867568074 0.8749524879 0.87873071 0.310400002 0.3207391676 0.3324358266 74
27 Medullo DNase None 0.864042242 0.8664238813 0.8692186275 0.3158031211 0.3271526972 0.3335229698 27
3 Chorion DNase None 0.896293207 0.9002669789 0.9008547544 0.3207430835 0.3384839394 0.3367010903 3
67 GM06990 DNase None 0.902955171 0.9063258167 0.9130242555 0.3256947324 0.3396407621 0.3519833383 67Codes
yanyu_lib.R
zoom_in <- function(melted_distance, data_none_ctcf){
sub.ctcf.first.ind <- melted_distance$Var1 %in% data_none_ctcf$row_num
sub.ctcf.second.ind <- melted_distance$Var2 %in% data_none_ctcf$row_num
sub.ctcf <- melted_distance[sub.ctcf.first.ind & sub.ctcf.second.ind, ]
sub.ctcf$Var1.sub <- rep(1:nrow(data_none_ctcf), nrow(data_none_ctcf))
sub.ctcf$Var2.sub <- rep(1:nrow(data_none_ctcf), each=nrow(data_none_ctcf))
data_none_ctcf$DeepSEA.ROC.AUC <- as.numeric(as.character(data_none_ctcf$DeepSEA.ROC.AUC))
data_none_ctcf$DeepSEA.PR.AUC <- as.numeric(as.character(data_none_ctcf$DeepSEA.PR.AUC))
data_none_ctcf$dist.mean <- aggregate(sub.ctcf$value, list(sub.ctcf$Var1), mean)$x
return(list(sub=sub.ctcf, ori=data_none_ctcf))
}
as_num <- function(x){
x$DeepSEA.ROC.AUC <- as.numeric(as.character(x$DeepSEA.ROC.AUC))
x$DeepSEA.PR.AUC <- as.numeric(as.character(x$DeepSEA.PR.AUC))
return(x)
}
## derived from http://www.cookbook-r.com/Graphs/Multiple_graphs_on_one_page_(ggplot2)/
multiplot <- function(..., plotlist=NULL, file, cols=1, layout=NULL) {
library(grid)
# Make a list from the ... arguments and plotlist
plots <- c(list(...), plotlist)
numPlots = length(plots)
# If layout is NULL, then use 'cols' to determine layout
if (is.null(layout)) {
# Make the panel
# ncol: Number of columns of plots
# nrow: Number of rows needed, calculated from # of cols
layout <- matrix(seq(1, cols * ceiling(numPlots/cols)),
ncol = cols, nrow = ceiling(numPlots/cols))
}
if (numPlots==1) {
print(plots[[1]])
} else {
# Set up the page
grid.newpage()
pushViewport(viewport(layout = grid.layout(nrow(layout), ncol(layout))))
# Make each plot, in the correct location
for (i in 1:numPlots) {
# Get the i,j matrix positions of the regions that contain this subplot
matchidx <- as.data.frame(which(layout == i, arr.ind = TRUE))
print(plots[[i]], vp = viewport(layout.pos.row = matchidx$row,
layout.pos.col = matchidx$col))
}
}
}
## endcompute_distance.py
import sys
if sys.argv[1] == '--help':
print('compute_distance.py [input_data_hdf5] [output_dir]')
sys.exit()
import h5py
sys.path.insert(0, '/project2/xinhe/yanyul/deep_variant/yanyu/DanQ/my_scripts')
from keras_mylib import my_load_model, load_standard_hdf5_data, my_build_model
import numpy as np
import ntpath
import os
import theano
import theano.tensor as T
from theano import function
print('Build computational graph in theano')
x = T.dmatrix('x')
y = T.dmatrix('y')
z = theano.tensor.dot(x, y)
dot = function([x, y], z)
print('Loading data')
X, y = load_standard_hdf5_data(sys.argv[1])
outname = sys.argv[2] + os.sep + ntpath.basename(sys.argv[1]) + '.y_disance.hdf5'
print('One & One')
one_one = dot(y.T, y)
f = h5py.File(outname, 'w')
f.create_dataset('one_one',data=one_one)
f.close()
print('Zero & Zero')
zero_zero = dot(1 - y.T, 1 - y)
f = h5py.File(outname, 'a')
f.create_dataset('zero_zero',data=zero_zero)
f.close()
print('Zero & One')
zero_one = dot(1 - y.T, y)
f = h5py.File(outname, 'a')
f.create_dataset('zero_one',data=zero_one)
f.close()
print('One & Zero')
one_zero = dot(y.T, 1 - y)
f = h5py.File(outname, 'a')
f.create_dataset('one_zero',data=one_zero)
f.close()Session information
sessionInfo()R version 3.3.3 (2017-03-06)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: OS X Yosemite 10.10.5
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] grid stats graphics grDevices utils datasets methods base
other attached packages:
[1] workflowr_0.4.0 plyr_1.8.4 rmarkdown_1.4 reshape2_1.4.2 h5_0.9.8 ggplot2_2.2.1
loaded via a namespace (and not attached):
[1] Rcpp_0.12.10 rstudioapi_0.6 knitr_1.15.1 magrittr_1.5 munsell_0.4.3 colorspace_1.3-2 stringr_1.2.0 tools_3.3.3 gtable_0.2.0 git2r_0.18.0 htmltools_0.3.5 yaml_2.1.14 lazyeval_0.2.0 assertthat_0.1 digest_0.6.12 rprojroot_1.2 tibble_1.2 evaluate_0.10 labeling_0.3 stringi_1.1.3 scales_0.4.1 backports_1.0.5 This R Markdown site was created with workflowr